US20020178135A1 - Image searching system and image searching method, and a recording medium storing an image searching program - Google Patents
Image searching system and image searching method, and a recording medium storing an image searching program Download PDFInfo
- Publication number
- US20020178135A1 US20020178135A1 US09/255,352 US25535299A US2002178135A1 US 20020178135 A1 US20020178135 A1 US 20020178135A1 US 25535299 A US25535299 A US 25535299A US 2002178135 A1 US2002178135 A1 US 2002178135A1
- Authority
- US
- United States
- Prior art keywords
- images
- controller
- image
- key
- database
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P11/00—Drugs for disorders of the respiratory system
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P13/00—Drugs for disorders of the urinary system
- A61P13/02—Drugs for disorders of the urinary system of urine or of the urinary tract, e.g. urine acidifiers
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P15/00—Drugs for genital or sexual disorders; Contraceptives
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P17/00—Drugs for dermatological disorders
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P27/00—Drugs for disorders of the senses
- A61P27/02—Ophthalmic agents
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P27/00—Drugs for disorders of the senses
- A61P27/16—Otologicals
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P29/00—Non-central analgesic, antipyretic or antiinflammatory agents, e.g. antirheumatic agents; Non-steroidal antiinflammatory drugs [NSAID]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P31/00—Antiinfectives, i.e. antibiotics, antiseptics, chemotherapeutics
- A61P31/04—Antibacterial agents
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
- A61P43/00—Drugs for specific purposes, not provided for in groups A61P1/00-A61P41/00
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10—TECHNICAL SUBJECTS COVERED BY FORMER USPC
- Y10S—TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10S707/00—Data processing: database and file management or data structures
- Y10S707/99931—Database or file accessing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10—TECHNICAL SUBJECTS COVERED BY FORMER USPC
- Y10S—TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10S707/00—Data processing: database and file management or data structures
- Y10S707/99931—Database or file accessing
- Y10S707/99933—Query processing, i.e. searching
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10—TECHNICAL SUBJECTS COVERED BY FORMER USPC
- Y10S—TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10S707/00—Data processing: database and file management or data structures
- Y10S707/99931—Database or file accessing
- Y10S707/99933—Query processing, i.e. searching
- Y10S707/99935—Query augmenting and refining, e.g. inexact access
Definitions
- the present invention relates generally to an image searching system for finding an image that resembles a specified image from among images managed by the image searching system where the search criteria are determined from the specified image.
- the present invention relates more specifically to an image searching system for finding an image based on a plurality of specified images.
- Images stored to a database for managing image data are typically stored with additional information that can be used to specify search conditions, including search keys, such as identifying keywords and image color keys, and image features, such as information on the shapes and colors appearing in the image.
- search keys such as identifying keywords and image color keys
- image features such as information on the shapes and colors appearing in the image.
- Image similarity in similar-image searching is determined based on such feature quantities as image shape, texture, and color. More specifically, the user specifies an image in the image database that is similar to a desired image to be searched for. Feature quantities are then extracted from the image selected by the user as the search key (referred to below as the “key image”), and the extracted feature quantities are then compared with the feature quantities registered in the image database. Image similarity indicative of the similarity between the key image and a found image is then calculated. Information on any stored image determined to resemble the key image based on this calculated image similarity is then returned as the search result.
- Japanese Patent Laid-Open Publication 7-65169 teaches one method proposing to solve this problem.
- a plurality of key images is specified, a specific area in each of the key images is selected, and the features to be used as search criteria are extracted from the selected areas.
- the search is then conducted using these features from plural selected areas in plural images.
- This method thus makes it possible to search for images having all of the features found in the specified areas of the plural key images. In other words, this method enables the user to specify a wide range of search conditions based on plural images.
- the present invention is therefore directed to an image searching system and an image searching method for reducing the user burden and improving search accuracy when a plurality of key images is specified and a desired image is searched for based on these selected key images.
- the present invention is further directed to a recording medium for storing an image searching program embodying the image searching method of the present invention.
- a first image searching system comprises: an image storage means for storing a plurality of target images for searching; a selection means for selecting a plurality of key images for determining the search criteria; a first feature quantity extraction means for extracting one or a plurality of feature quantities from each of the plurality of key images selected by the selection means; a second feature quantity extraction means for extracting one or a plurality of feature quantities from a target image; a feature quantity identification means for comparing a feature quantity extracted by the first feature quantity extraction means with the same type of feature quantity in each of the plurality of key images to determine a feature quantity to be used for searching; a degree of similarity calculating means for calculating a degree of similarity between a target image and key image using the feature quantity identified by the feature quantity identification means for searching; and an extraction means for extracting as an image similar to a key image any target image for which the degree of similarity calculated by the degree of similarity calculating means exceeds a specific value.
- this first image searching system compares features extracted from plural key images to determine the features to be used as search criteria, and then calculates image similarity using the identified features (feature quantities).
- the feature quantity identification means preferably defines as a feature quantity type to be used for searching a feature quantity type having an approximately equal value in each of the compared key images.
- a second image searching system comprises: an image storage means for storing a plurality of target images for searching; a selection means for selecting a plurality of key images for determining the search criteria; a first feature quantity extraction means for extracting a feature quantity from each of the plurality of key images selected by the selection means; a second feature quantity extraction means for extracting a feature quantity from a target image; a first degree of similarity calculating means for calculating a first degree of similarity indicative of the similarity between a target image and all of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction means; a second degree of similarity calculating means for calculating a second degree of similarity indicative of the similarity between a target image and at least one of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction means; a third degree of similarity calculating means for calculating a third degree of similarity from the first and second degrees of similarity where the first degree of similarity is weighted greater than the
- a first image searching method searches in a plurality of stored target images for an image that is similar to a key image based on the key image.
- This first image searching method comprises: a selection step for selecting a plurality of key images; a first feature quantity extraction step for extracting one or a plurality of feature quantity types from each of the plurality of selected key images; a second feature quantity extraction step for extracting one or a plurality of feature quantity types from the target images; a feature quantity identification step for comparing a feature quantity extracted by the first feature quantity extraction step with the same type of feature quantity in each of the plurality of key images to determine a feature quantity type to be used for searching; a degree of similarity calculating step for calculating a degree of similarity between a target image and key image using the feature quantity type identified for searching; and an extraction step for extracting as an image similar to a key image any target image for which the calculated degree of similarity exceeds a specific value.
- a second image searching method for searching in a plurality of stored target images for an image that is similar to a key image based on said key image comprises: a selection step for selecting a plurality of key images; a first feature quantity extraction step for extracting a feature quantity from each of the plurality of selected key images; a second feature quantity extraction step for extracting a feature quantity from a target image; a first degree of similarity calculating step for calculating a first degree of similarity indicative of the similarity between a target image and all of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a second degree of similarity calculating means for calculating a second degree of similarity indicative of the similarity between a target image and at least one of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a third degree of similarity calculating step for calculating a third degree of similarity from the first and second degrees of similarity where the first degree of similarity is weighted greater than the second degree of similar
- a first recording medium is a computer-readable recording medium for recording an image searching program where the image searching program embodies a method for searching in a plurality of stored target images for an image that is similar to a key image based on said key image.
- Specific steps in this image searching program include: a selection step for selecting a plurality of key images; a first feature quantity extraction step for extracting one or a plurality of feature quantity types from each of the plurality of selected key images; a second feature quantity extraction step for extracting one or a plurality of feature quantity types from the target images; a feature quantity identification step for comparing a feature quantity extracted by the first feature quantity extraction step with the same type of feature quantity in each of the plurality of key images to determine a feature quantity type to be used for searching; a degree of similarity calculating step for calculating a degree of similarity between a target image and key image using the feature quantity type identified for searching; and an extraction step for extracting as an image similar to a key image any target image for which the calculated degree of similarity exceeds a specific value.
- a second recording medium is also a computer-readable recording medium for recording an image searching program.
- Specific steps in this image searching program include: a selection step for selecting a plurality of key images; a first feature quantity extraction step for extracting a feature quantity from each of the plurality of selected key images; a second feature quantity extraction step for extracting a feature quantity from a target image; a first degree of similarity calculating step for calculating a first degree of similarity indicative of the similarity between a target image and all of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a second degree of similarity calculating means for calculating a second degree of similarity indicative of the similarity between a target image and at least one of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a second degree of similarity calculating means for calculating a second degree of similarity indicative of the similarity between a target image and at least one of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction
- FIG. 1 is a typical overview of an image searching system according to a preferred embodiment of the present invention
- FIG. 2 is a block diagram of primarily the control device of the image searching system shown in FIG. 1;
- FIG. 3 shows a typical image database table structure
- FIG. 4 is a flow chart of the main loop of the image searching method embodied in the image searching system shown in FIG. 1;
- FIG. 5 is a flow chart of the image registration process shown as step S 3 in FIG. 4;
- FIG. 6 is a flow chart of the similar image searching process shown as step S 4 in FIG. 4;
- FIG. 7 is a flow chart of the search feature compilation process shown as step S 43 in FIG. 6;
- FIG. 8 is a data table used to describe a method of calculating common features
- FIG. 9 is a data table of sample feature quantities extracted from a key image
- FIG. 10 is a data table of sample feature quantities extracted from a target image
- FIG. 11 is a data table used to describe similarity calculation based on common feature quantities
- FIG. 12 is a flowchart of the similar image searching process shown as step S 4 in FIG. 4, that is executed according to a second preferred embodiment of the present invention.
- FIG. 13 is a flowchart of the search feature compilation process shown as step S 43 in FIG. 6, that is executed according to the second preferred embodiment of the present invention.
- FIG. 14 is a data table used to describe similarity calculation based on all feature quantities
- FIG. 15 is a data table used to describe similarity calculation based on a degree of similarity based on common feature quantities and a degree of similarity based on all feature quantities.
- An image searching system searches for an image similar to common features found in a plurality of specified key images.
- the image searching system extracts one or a plurality of feature quantities from the plural key images and compares similar feature quantities in the key images to determine the types of feature quantities to use for searching.
- the degree of similarity is then calculated using the specified feature quantity types, and an image search is then conducted based on this degree of similarity. It is therefore possible to search for images having features common to a plurality of key images.
- the image searching system of the present invention can alternatively calculate the degree of similarity using both a degree of similarity indicative of the resemblance to all of the plural key images, and a degree of similarity indicative of the resemblance to one or some larger subset of the plural key images. This method makes it possible to expand the range of the search, and thus improve search accuracy.
- FIG. 1 shows the basic configuration of an image searching system, referred to as the “system” below, according to a preferred embodiment of the present invention.
- this system is built around a control device 1 for controlling the overall system and comprising a central processing unit (CPU).
- CPU central processing unit
- a display 2 for displaying text and images, as well as control prompts and other interface elements; a keyboard 3 and mouse 4 for data entry and operating the system; a floppy disk drive 5 b in which a floppy disk 5 a can be used and a hard disk drive 6 for reading and writing data to a data storage medium; a printer 7 for printing text and images to hard copy; a scanner 8 for capturing image data; a CD-ROM drive 9 b for accessing data stored to a CD-ROM 9 a ; a speaker 10 for audio output; and a microphone 11 for audio input.
- FIG. 2 is a block diagram of the image searching system shown in FIG. 1.
- a clock 202 for generating the reference clock required to regulate system operations.
- a ROM 203 for storing an application program for controlling the image searching system, and a RAM 204 for temporarily storing data and programs used by the CPU 201 for system control, are connected to the CPU 201 via a data bus 220 .
- Circuits connected to the CPU 201 via the data bus 220 include: a display control circuit 205 for controlling the display 2 to display text and graphics; a keyboard control circuit 206 for controlling transfer of input from the keyboard 3 ; a mouse control circuit 207 for controlling transfer of input from the mouse 4 ; a floppy disk drive control circuit 208 for controlling the floppy disk drive 5 b ; a hard disk drive control circuit 209 for controlling the hard disk drive 6 ; a printer control circuit 210 for controlling output to the printer 7 ; a scanner control circuit 211 for controlling the scanner 8 ; a CD-ROM drive control circuit 212 for controlling the CD-ROM drive 9 b ; a speaker control circuit 213 for controlling the speaker 10 ; and a microphone control circuit 214 for controlling the microphone 11 .
- An expansion slot 215 for connecting different types of system expansion boards is also connected via the data bus 220 .
- a SCSI board for example, can be connected to the expansion slot 215 and used for connecting a floppy disk drive 5 b , hard disk drive 6 , scanner 8 , CD-ROM drive 9 b , or other device.
- floppy disk 5 a and hard disk drive 6 are the preferred image data storage means in this exemplary system, the invention shall obviously not be limited thereto as any other data storage medium that the system can use can be used for image data storage, including magneto-optical (MO) disks [and DVD disks].
- MO magneto-optical
- a scanner 8 is used in this exemplary system for image data capture and input, other data input devices can also be used, including video still cameras and digital cameras.
- printer 7 is used as an output device
- a digital photocopier or other type of hard copy output device can also be used.
- a program embodying the control method of the present invention is also stored to ROM 203 in this preferred embodiment.
- This control program can be alternatively stored in whole or in part to another data storage medium, including a floppy disk Sa, hard disk drive 6 , or CD-ROM 9 a .
- the required data and program content can be read from the data storage medium as needed, buffered to RAM 204 , and executed from RAM 204 .
- an image searching system also has an image database and a color space table.
- the image database contains both the image data and additional information used for the search keys.
- the color space table is referenced when extracting color features from an image.
- Both the image database and color space table are logic objects stored on the hard disk drive 6 or other data storage medium.
- FIG. 3 An exemplary image database 50 in the present system is shown in FIG. 3.
- this image database 50 manages the image data to be searched, and various columns of image attributes that are used as search parameters for finding desired image data.
- image attributes include keywords, a color key, and other search keys, and various feature quantities indicative of specific features in the image data. More specifically, these feature quantities include a color feature indicative of features relating to image color, a shape feature indicative of shapes in the image, and a texture feature indicative of texture patterns. The similarity between images is calculated based on this feature information.
- FIG. 4 is a flow chart of the main loop of a control program run by the CPU 201 in the image searching system shown in FIG. 1.
- the procedure starts by initializing the various flags and variables used in the rest of the procedure, and to display the default menus (S 1 ).
- An exemplary default menu screen displays various icons representing and used for selecting specific processes such that when a user selects one of the icons, the corresponding process is executed. It should be further noted that the system of this preferred embodiment enables the operator to select various processes and enter information using the keyboard 3 and mouse 4 from the default menu screen and other screens that are displayed on the display 2 .
- decision diamond S 2 detects whether the user has selected a menu item from the default menu screen (S 2 ).
- step S 2 If in step S 2 the user selects “register image,” for example, the image registration process (S 3 ) is selected and run, and the procedure then advances to step S 6 .
- This image registration process (S 3 ) is run to store image data and related feature information in the image database 50 , and is described in further detail below with reference to the flow chart in FIG. 5.
- step S 2 If in step S 2 the user selects “search for similar image,” for example, the similar image searching process (S 4 ) is selected and run, and the procedure then advances to step S 6 .
- This similar image searching process (S 4 ) is run to search in the image database 50 for an image similar to a selected key image, and is described in further detail below with reference to the flow chart in FIG. 6.
- step S 2 If in step S 2 the user selects another menu item, the corresponding process (S 5 ) is selected and run, and the procedure then advances to step S 6 .
- step S 6 If no menu item is selected in S 2 , the procedure advances directly to step S 6 . After the procedure defined for step S 6 is run and completed, the main loop returns to step S 2 and repeats.
- step S 3 in FIG. 4 The image registration process shown as step S 3 in FIG. 4 is described next in detail below with reference to the flow chart in FIG. 5. This process extracts feature quantities from a user-specified image to be stored in the image database, and then stores the specified image and extracted features to the image database 50 .
- This routine starts by reading the user-specified image to be stored to the image database 50 (S 31 ).
- a feature extraction process is then run (S 32 ) to extract feature quantities from the specified image.
- the extracted feature quantities include color, shape, and texture, but the invention shall obviously not be limited thereto.
- the extracted feature quantities are then related* to the image data of the specified image (S 33 ), and finally registered in the image database 50 together with the related specified image data (S 34 ).
- a file name, image size information, and other image attributes are also normally stored with or related to the image data.
- the procedure then returns to the main loop (step S 6 in FIG. 4).
- common feature quantities are extracted from a plurality of key images. Upon evaluating a similarity of a target image to a key image, only these common feature quantities are used for calculating a degree of similarity of the target image to a key image. In other words, feature quantities other than the common feature quantities are not used for calculation of the degree of similarity.
- step 4 in FIG. 4 The similar image searching processing according to first embodiment (shown as step 4 in FIG. 4) is described below with reference to a flow chart shown in FIG. 6.
- predetermined feature quantities are extracted from each of plural key images specified by a user and feature quantities indicative of features common to the plural key images are selected as common feature quantities among the extracted feature quantities.
- the searching processing is performed to find out an image similar to a key image among images registered in the image database 50 based on the common feature quantities.
- a plurality of key images specified by a user are read (step S 41 ).
- Feature quantities of each key image are extracted from the image database 50 (step S 42 ).
- feature quantities of plural kinds such as color, shape, texture and so on are obtained from each key image.
- a search feature quantity compilation processing is performed to calculate common feature quantities for searching a similar image by comparing the feature quantities of respective key images with each other (step S 43 ). This search feature quantity compilation processing will be described later.
- step S 43 the common feature quantities calculated in step S 43 are compared with feature quantities of each image registered in the image database, and the degree of similarity between the key images and the presently selected image is calculated (from step S 44 to step S 46 ).
- step S 44 it is determined if there is an image in the image database 50 for which the degree of similarity to the key images has not yet been calculated (herein below referred to as a “target image”) (step S 44 ). If such a target image is present in the image database 50 , the feature quantities for that target image are read from the image database 50 (step S 45 ). Then, a degree of similarity indicative of a similarity between the target image and each key image is calculated based on the common feature quantities of the key images and the feature quantities of the key images and the feature quantities of the target image (step S 46 ) and the processing loops back to step S 43 . This calculation processing of the degree of similarity will be described later. These steps S 43 to S 45 are repeated until the similarity calculation has been completed for all target images in the image database.
- step S 47 an output processing of search results is performed (step S 47 ) and the processing returns to the main routine.
- the output processing of the search results it is determined that those target images for which the calculated degree of similarity exceeds a predetermined threshold value are similar to the key images.
- the target images determined to be similar are then displayed on screen in turn from the highest degree of similarity.
- a predetermined number of target images can be displayed in a descending order from the highest degree of similarity.
- the search results can be displayed by simply outputting the file name of the target image data, preferably with related attributes such as its location in the database.
- the target images tested by the search loop from step S 44 to step S 46 shall also not be limited to all of the images stored in the image database 50 . For instance, only images preselected based on predetermined conditions can be tested.
- the search feature quantity compilation processing (step S 43 in FIG. 6) is described next with reference to a flow chart shown in FIG. 7.
- This processing calculates the common feature quantities to be used upon searching a target image which is similar in common to a plurality of key images. That is, the common feature quantities indicative of features in common to a plurality of key images are obtained by comparing feature quantities of the same kind among the plurality of key images and choosing feature quantities of the same kind having values being approximate to each other as the common feature quantities. This processing is described in further detail below.
- the processing starts by calculating, for every kind of the feature quantity, differences among the feature quantities of the key images (step S 431 ).
- a mean value of the differences obtained in step S 431 is calculated for every kind of the feature quantity (S 432 ).
- FIG. 8 This figure shows an example obtained for three key images 1 to 3 each having five feature quantities 1 to 5 as shown in FIG. 9.
- the mean value of the differences with respect to the same feature quantity indicates the smaller the mean value the greater the image similarity with respect to that feature quantity.
- every feature quantity or the kind of the feature quantity having a mean value equal to or less than a predetermined threshold is selected as a common feature quantity (step S 433 ). That is, when the mean feature quantity difference is less than the threshold value, that feature quantity is determined to be common to the plural key images.
- this threshold value is set to 0.1 and feature quantity 2 and feature quantity 3 are thus selected as common feature quantities.
- the feature quantities or kinds of feature quantities to be used for searching are determined by comparing feature quantities of the same kind in the key images.
- a specific flag is set for every feature quantity selected as a common feature quantity (step S 434 ).
- step S 46 in FIG. 6 the processing for calculating the degree of similarity (step S 46 in FIG. 6) is described below. Based on the common feature quantities extracted from the key images and a target image, this processing calculates the degree of similarity of the target image to all key images or either one of chem.
- step S 43 only feature quantities selected as common feature quantities in step S 43 are used for calculating the degree of similarity of the target image to the key image.
- FIG. 11 shows an example of the calculation for the degree of similarity.
- the distance between each key image and the target image is calculated based on the values of feature quantity 2 and feature quantity 3 selected as the common feature quantities and the degree of similarity is obtained based on the calculated distances.
- the values shown in FIG. 11 are obtained from the following equations (1) to (3).
- the degree of similarity thus obtained indicates that the image similarity becomes high as the calculated value of the degree of similarity becomes large.
- the degree of similarity between a target image and key images is calculated for each of the key images based on feature quantities of all kinds. From the degrees of similarity calculated for each key image, the highest degree of similarity is then selected as the degree of similarity of the target image to the corresponding key image. In this case, images similar to at least one of the plural key images are searched. That is, the image search is performed under an OR logic condition and, accordingly, it becomes possible to broaden the ranges of searching conditions by increasing the number of key images.
- Second embodiment performs a processing shown in FIG. 12 and FIG. 13 instead of FIGS. 6 and 7 of first embodiment.
- a plurality of key images specified by a user are read at first (S 141 ).
- Feature quantities are obtained from the image database 50 for each key image (S 142 ). Namely, feature quantities of plural kinds such as color, shape, texture and the like are obtained form each key image.
- step S 142 the feature quantities of each key image obtained in step S 142 are compared with feature quantities of each target image registered in the image database 50 , the similarity between each key image and the target image is determined and thereby, target images being similar to respective key images are searched (steps from S 143 to S 145 ).
- step S 143 it is determined whether or not there is a target image in the image database for which the degree of similarity to each key image has not yet calculated. If such a target image is present in the image database 50 , the feature quantities for that target image are read from the image database (step S 144 ). Then, calculation of the degree of similarity indicative of similarity between images is performed based on the feature quantities of each key image and those of the target image (step S 144 ). After this calculation, the processing loops back to step S 143 . This calculation processing of the degree of similarity (step S 145 ) will be described later. These steps S 143 to S 145 are repeated until the similarity calculation has been completed for all target images in the image database 50 .
- step S 147 an output processing of search results is performed (step S 147 ) and then, the processing returns to the main routine.
- this output processing of search results (step S 146 ) based on the degrees of similarity calculated for all target images in the image database 50 , the key image having a degree of similarity higher than a predetermined threshold value is determined to be similar to the corresponding key image.
- the target images determined to be similar are then display on screen in sequence from the highest degree of similarity.
- a predetermined number of target images can be displayed in a descending order from the highest degree of similarity.
- the search results can be displayed by simply outputting the file name of the target image data, preferably with related attributes such as its location in the database.
- the target images tested by the search loop from step S 144 to step S 146 shall also not be limited to all of the images stored in the image database 50 . For instance, only images preselected based on predetermined conditions can be tested.
- a distance between a feature quantity of the key image and that of the same kind of the target image is calculated for each of all feature quantities, and a distance between the key images and the target image is calculated based on the distances of respective feature quantities.
- step S 1451 Based on the distance between images, the degree of similarity is calculated (step S 1451 ). More specifically, the values shown in FIG. 14 are obtained from the following equations (4) to (6).
- step S 146 The output processing of search results mentioned above with respect to step S 146 is performed based on the degrees of similarity determined in step S 1452 .
- This method combines the degrees of similarity based on the common feature quantities calculated according to first embodiment (a) and those based on all feature quantities calculated according to second embodiment (b), to obtain final degrees of similarity.
- the degree of similarity based on common feature quantities is weighted and the weighted degree of similarity and the degree of similarity based on all feature quantities are added to obtain a mean value thereof.
- This mean value is assigned as the degree of similarity of the target image to the key image.
- FIG. 15 An example of the calculation of the degree of similarity is shown in FIG. 15.
- the weight for the degree of similarity a calculated in FIG. 11 is set larger than that for the degree of similarity calculated in FIG. 14.
- a weighted mean value is than calculated to obtain the final degree of similarity c.
- This final degree of similarity c is obtained according to the next equation (7).
- Degree of similarity ⁇ (degree of similarity a ⁇ weight wa )+(degree of similarity b ⁇ weight wb ) ⁇ /( wa+wb ) (7)
- feature quantities are extracted from a plurality of key images, when specified and those common to the plural key images are selected to determine similarity between images using these common feature quantities.
- the common feature quantities are automatically determined from among plural key images and accordingly, the user burden can be reduced.
- search for images having features common to plural key images This enables to narrow the ranges of respective searching conditions by specifying plural key images.
- the search can be similarly expanded to broader ranges of searching conditions by considering the degree of similarity indicating similarity of an image to at least one of plural key images. Search accuracy can be thus improved.
- a first image searching system, a first image searching method, and a first recording medium storing an image searching program embodying the first image searching method of the invention determines the feature quantities to be used for image searching based on the feature quantities extracted from one or more key images, and calculates a degree of similarity using only the feature quantities selected as search criteria. It is therefore possible to search for images with features common to those of the selected key images.
- the search conditions can thus be restricted by specifying plural key images, and search accuracy can thus be improved.
- the user burden is also reduced because feature quantities are automatically extracted from the specified key images, and similar images are searched for based on the extracted feature quantities.
- a second image searching system, a second image searching method, and a second recording medium storing an image searching program embodying the second image searching method of the invention calculates the degree of similarity used for searching by combining a degree of similarity indicative of similarity to all of the plurality of key image, and a degree of similarity indicative of similarity to any one of the plurality of key image. It is therefore possible to set a wide range of search criteria by selecting a plurality of key images, and further improve the accuracy of similar image searches.
Abstract
An image searching system, an image searching method, and a medium storing an image searching program for improving search accuracy while reducing the user burden when searching for similar images based on a plurality of key images are disclosed.
A plurality of feature quantity types is extracted from each of plural key images (S42), and feature quantities of the same type are compared between the key images to select those feature quantities having approximately equal values as the key image feature quantities to be used as search criteria (S43). Feature quantities in the target images are then extracted (S45) and a degree of similarity between the target image and key images is calculated based on a key image feature quantity and the same feature quantity in the target image (S46) Based on the resulting degree of similarity, target images having a high similarity to the key image are output as the search result (S47).
Description
- 1. Field of the Invention
- The present invention relates generally to an image searching system for finding an image that resembles a specified image from among images managed by the image searching system where the search criteria are determined from the specified image. The present invention relates more specifically to an image searching system for finding an image based on a plurality of specified images.
- 2. Description of the Related Art
- Images stored to a database for managing image data are typically stored with additional information that can be used to specify search conditions, including search keys, such as identifying keywords and image color keys, and image features, such as information on the shapes and colors appearing in the image. Using this common type of image database, it is possible to search the stored images to find an image similar to a user-specified image.
- Image similarity in similar-image searching is determined based on such feature quantities as image shape, texture, and color. More specifically, the user specifies an image in the image database that is similar to a desired image to be searched for. Feature quantities are then extracted from the image selected by the user as the search key (referred to below as the “key image”), and the extracted feature quantities are then compared with the feature quantities registered in the image database. Image similarity indicative of the similarity between the key image and a found image is then calculated. Information on any stored image determined to resemble the key image based on this calculated image similarity is then returned as the search result.
- Usually there is only one key image selected by the user, and it is often impossible to find the one image sought by the user based on the feature quantities defined from a single key image. Search accuracy is therefore often low. This can happen, for example, when there are plural features in a single image and the user focuses on only one feature, that is, when the user's search criteria are narrower than the criteria that can be extracted from the key image, as well as when numerous search conditions are specified to broaden the range of the search.
- Japanese Patent Laid-Open Publication 7-65169 teaches one method proposing to solve this problem. According to the taught method, a plurality of key images is specified, a specific area in each of the key images is selected, and the features to be used as search criteria are extracted from the selected areas. The search is then conducted using these features from plural selected areas in plural images. This method thus makes it possible to search for images having all of the features found in the specified areas of the plural key images. In other words, this method enables the user to specify a wide range of search conditions based on plural images.
- The drawback to said taught method is that the user is required to specify in each of the selected key images those features that are to be used for searching, and this task becomes increasingly difficult and time-consuming as the number of key images increases and the variety of selected features increases.
- The present invention is therefore directed to an image searching system and an image searching method for reducing the user burden and improving search accuracy when a plurality of key images is specified and a desired image is searched for based on these selected key images.
- The present invention is further directed to a recording medium for storing an image searching program embodying the image searching method of the present invention.
- A first image searching system according to the present invention comprises: an image storage means for storing a plurality of target images for searching; a selection means for selecting a plurality of key images for determining the search criteria; a first feature quantity extraction means for extracting one or a plurality of feature quantities from each of the plurality of key images selected by the selection means; a second feature quantity extraction means for extracting one or a plurality of feature quantities from a target image; a feature quantity identification means for comparing a feature quantity extracted by the first feature quantity extraction means with the same type of feature quantity in each of the plurality of key images to determine a feature quantity to be used for searching; a degree of similarity calculating means for calculating a degree of similarity between a target image and key image using the feature quantity identified by the feature quantity identification means for searching; and an extraction means for extracting as an image similar to a key image any target image for which the degree of similarity calculated by the degree of similarity calculating means exceeds a specific value.
- Thus comprised, this first image searching system compares features extracted from plural key images to determine the features to be used as search criteria, and then calculates image similarity using the identified features (feature quantities).
- In this first image searching system, the feature quantity identification means preferably defines as a feature quantity type to be used for searching a feature quantity type having an approximately equal value in each of the compared key images.
- A second image searching system according to the present invention comprises: an image storage means for storing a plurality of target images for searching; a selection means for selecting a plurality of key images for determining the search criteria; a first feature quantity extraction means for extracting a feature quantity from each of the plurality of key images selected by the selection means; a second feature quantity extraction means for extracting a feature quantity from a target image; a first degree of similarity calculating means for calculating a first degree of similarity indicative of the similarity between a target image and all of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction means; a second degree of similarity calculating means for calculating a second degree of similarity indicative of the similarity between a target image and at least one of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction means; a third degree of similarity calculating means for calculating a third degree of similarity from the first and second degrees of similarity where the first degree of similarity is weighted greater than the second degree of similarity; and an extraction means for extracting as an image similar to a key image any target image for which the third degree of similarity calculated by the third degree of similarity calculating means exceeds a specific value. The third degree of similarity in this case is calculated with greater weight assigned to the first degree of similarity than to the second degree of similarity.
- A first image searching method according to the present invention searches in a plurality of stored target images for an image that is similar to a key image based on the key image. This first image searching method comprises: a selection step for selecting a plurality of key images; a first feature quantity extraction step for extracting one or a plurality of feature quantity types from each of the plurality of selected key images; a second feature quantity extraction step for extracting one or a plurality of feature quantity types from the target images; a feature quantity identification step for comparing a feature quantity extracted by the first feature quantity extraction step with the same type of feature quantity in each of the plurality of key images to determine a feature quantity type to be used for searching; a degree of similarity calculating step for calculating a degree of similarity between a target image and key image using the feature quantity type identified for searching; and an extraction step for extracting as an image similar to a key image any target image for which the calculated degree of similarity exceeds a specific value.
- A second image searching method for searching in a plurality of stored target images for an image that is similar to a key image based on said key image comprises: a selection step for selecting a plurality of key images; a first feature quantity extraction step for extracting a feature quantity from each of the plurality of selected key images; a second feature quantity extraction step for extracting a feature quantity from a target image; a first degree of similarity calculating step for calculating a first degree of similarity indicative of the similarity between a target image and all of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a second degree of similarity calculating means for calculating a second degree of similarity indicative of the similarity between a target image and at least one of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a third degree of similarity calculating step for calculating a third degree of similarity from the first and second degrees of similarity where the first degree of similarity is weighted greater than the second degree of similarity; and an extraction step for extracting as an image similar to a key image any target image for which the third degree of similarity exceeds a specific value.
- A first recording medium according to the present invention is a computer-readable recording medium for recording an image searching program where the image searching program embodies a method for searching in a plurality of stored target images for an image that is similar to a key image based on said key image. Specific steps in this image searching program include: a selection step for selecting a plurality of key images; a first feature quantity extraction step for extracting one or a plurality of feature quantity types from each of the plurality of selected key images; a second feature quantity extraction step for extracting one or a plurality of feature quantity types from the target images; a feature quantity identification step for comparing a feature quantity extracted by the first feature quantity extraction step with the same type of feature quantity in each of the plurality of key images to determine a feature quantity type to be used for searching; a degree of similarity calculating step for calculating a degree of similarity between a target image and key image using the feature quantity type identified for searching; and an extraction step for extracting as an image similar to a key image any target image for which the calculated degree of similarity exceeds a specific value.
- A second recording medium according to the present invention is also a computer-readable recording medium for recording an image searching program. Specific steps in this image searching program include: a selection step for selecting a plurality of key images; a first feature quantity extraction step for extracting a feature quantity from each of the plurality of selected key images; a second feature quantity extraction step for extracting a feature quantity from a target image; a first degree of similarity calculating step for calculating a first degree of similarity indicative of the similarity between a target image and all of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a second degree of similarity calculating means for calculating a second degree of similarity indicative of the similarity between a target image and at least one of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a second degree of similarity calculating means for calculating a second degree of similarity indicative of the similarity between a target image and at least one of the plurality of key images using the feature quantities extracted by the first and second feature quantity extraction steps; a third degree of similarity calculating step for calculating a third degree of similarity from the first and second degrees of similarity where the first degree of similarity is weighted greater than the second degree of similarity; and an extraction step for extracting as an image similar to a key image any target image for which the third degree of similarity exceeds a specific value.
- These and other objects and features of the present invention will be readily understood from the following detailed description taken in conjunction with preferred embodiments thereof with reference to the accompanying drawings, in which like parts are designated by like reference numerals and in which:
- FIG. 1 is a typical overview of an image searching system according to a preferred embodiment of the present invention;
- FIG. 2 is a block diagram of primarily the control device of the image searching system shown in FIG. 1;
- FIG. 3 shows a typical image database table structure;
- FIG. 4 is a flow chart of the main loop of the image searching method embodied in the image searching system shown in FIG. 1;
- FIG. 5 is a flow chart of the image registration process shown as step S3 in FIG. 4;
- FIG. 6 is a flow chart of the similar image searching process shown as step S4 in FIG. 4;
- FIG. 7 is a flow chart of the search feature compilation process shown as step S43 in FIG. 6;
- FIG. 8 is a data table used to describe a method of calculating common features;
- FIG. 9 is a data table of sample feature quantities extracted from a key image;
- FIG. 10 is a data table of sample feature quantities extracted from a target image;
- FIG. 11 is a data table used to describe similarity calculation based on common feature quantities;
- FIG. 12 is a flowchart of the similar image searching process shown as step S4 in FIG. 4, that is executed according to a second preferred embodiment of the present invention; and
- FIG. 13 is a flowchart of the search feature compilation process shown as step S43 in FIG. 6, that is executed according to the second preferred embodiment of the present invention.
- FIG. 14 is a data table used to describe similarity calculation based on all feature quantities;
- FIG. 15 is a data table used to describe similarity calculation based on a degree of similarity based on common feature quantities and a degree of similarity based on all feature quantities.
- The preferred embodiments of an image searching system according to the present invention are described below with reference to the accompanying figures. An image searching system according to the present invention searches for an image similar to common features found in a plurality of specified key images. The image searching system extracts one or a plurality of feature quantities from the plural key images and compares similar feature quantities in the key images to determine the types of feature quantities to use for searching. The degree of similarity is then calculated using the specified feature quantity types, and an image search is then conducted based on this degree of similarity. It is therefore possible to search for images having features common to a plurality of key images.
- The image searching system of the present invention can alternatively calculate the degree of similarity using both a degree of similarity indicative of the resemblance to all of the plural key images, and a degree of similarity indicative of the resemblance to one or some larger subset of the plural key images. This method makes it possible to expand the range of the search, and thus improve search accuracy.
- Overall Configuration of the Image Searching System
- FIG. 1 shows the basic configuration of an image searching system, referred to as the “system” below, according to a preferred embodiment of the present invention. As will be known from FIG. 1, this system is built around a
control device 1 for controlling the overall system and comprising a central processing unit (CPU). Connected to thisexemplary control device 1 are: adisplay 2 for displaying text and images, as well as control prompts and other interface elements; akeyboard 3 andmouse 4 for data entry and operating the system; afloppy disk drive 5 b in which afloppy disk 5 a can be used and ahard disk drive 6 for reading and writing data to a data storage medium; a printer 7 for printing text and images to hard copy; ascanner 8 for capturing image data; a CD-ROM drive 9 b for accessing data stored to a CD-ROM 9 a; a speaker 10 for audio output; and a microphone 11 for audio input. - FIG. 2 is a block diagram of the image searching system shown in FIG. 1.
- Connected directly to the
CPU 201 is aclock 202 for generating the reference clock required to regulate system operations. AROM 203 for storing an application program for controlling the image searching system, and aRAM 204 for temporarily storing data and programs used by theCPU 201 for system control, are connected to theCPU 201 via adata bus 220. Circuits connected to theCPU 201 via thedata bus 220 include: adisplay control circuit 205 for controlling thedisplay 2 to display text and graphics; akeyboard control circuit 206 for controlling transfer of input from thekeyboard 3; amouse control circuit 207 for controlling transfer of input from themouse 4; a floppy diskdrive control circuit 208 for controlling thefloppy disk drive 5 b; a hard diskdrive control circuit 209 for controlling thehard disk drive 6; aprinter control circuit 210 for controlling output to the printer 7; ascanner control circuit 211 for controlling thescanner 8; a CD-ROMdrive control circuit 212 for controlling the CD-ROM drive 9 b; aspeaker control circuit 213 for controlling the speaker 10; and amicrophone control circuit 214 for controlling the microphone 11. Anexpansion slot 215 for connecting different types of system expansion boards is also connected via thedata bus 220. - It should be noted that a SCSI board, for example, can be connected to the
expansion slot 215 and used for connecting afloppy disk drive 5 b,hard disk drive 6,scanner 8, CD-ROM drive 9 b, or other device. - It should also be noted that while a
floppy disk 5 a andhard disk drive 6 are the preferred image data storage means in this exemplary system, the invention shall obviously not be limited thereto as any other data storage medium that the system can use can be used for image data storage, including magneto-optical (MO) disks [and DVD disks]. - Furthermore, while a
scanner 8 is used in this exemplary system for image data capture and input, other data input devices can also be used, including video still cameras and digital cameras. - In addition, while a printer7 is used as an output device, a digital photocopier or other type of hard copy output device can also be used.
- A program embodying the control method of the present invention is also stored to
ROM 203 in this preferred embodiment. This control program, however, can be alternatively stored in whole or in part to another data storage medium, including a floppy disk Sa,hard disk drive 6, or CD-ROM 9 a. In this case, the required data and program content can be read from the data storage medium as needed, buffered toRAM 204, and executed fromRAM 204. - Image Database
- To store and manage image data, an image searching system according to this preferred embodiment also has an image database and a color space table. The image database contains both the image data and additional information used for the search keys. The color space table is referenced when extracting color features from an image. Both the image database and color space table are logic objects stored on the
hard disk drive 6 or other data storage medium. - An
exemplary image database 50 in the present system is shown in FIG. 3. As will be known from FIG. 3, thisimage database 50 manages the image data to be searched, and various columns of image attributes that are used as search parameters for finding desired image data. These image attributes include keywords, a color key, and other search keys, and various feature quantities indicative of specific features in the image data. More specifically, these feature quantities include a color feature indicative of features relating to image color, a shape feature indicative of shapes in the image, and a texture feature indicative of texture patterns. The similarity between images is calculated based on this feature information. - Controlling the Image Searching System
- A specific method of controlling the above-described image searching system is described next below with reference to the accompanying flow charts.
- Main Loop of the Control Program
- FIG. 4 is a flow chart of the main loop of a control program run by the
CPU 201 in the image searching system shown in FIG. 1. - The procedure starts by initializing the various flags and variables used in the rest of the procedure, and to display the default menus (S1). An exemplary default menu screen displays various icons representing and used for selecting specific processes such that when a user selects one of the icons, the corresponding process is executed. It should be further noted that the system of this preferred embodiment enables the operator to select various processes and enter information using the
keyboard 3 andmouse 4 from the default menu screen and other screens that are displayed on thedisplay 2. After initialization in step S1, decision diamond S2 detects whether the user has selected a menu item from the default menu screen (S2). - If in step S2 the user selects “register image,” for example, the image registration process (S3) is selected and run, and the procedure then advances to step S6. This image registration process (S3) is run to store image data and related feature information in the
image database 50, and is described in further detail below with reference to the flow chart in FIG. 5. - If in step S2 the user selects “search for similar image,” for example, the similar image searching process (S4) is selected and run, and the procedure then advances to step S6. This similar image searching process (S4) is run to search in the
image database 50 for an image similar to a selected key image, and is described in further detail below with reference to the flow chart in FIG. 6. - If in step S2 the user selects another menu item, the corresponding process (S5) is selected and run, and the procedure then advances to step S6.
- If no menu item is selected in S2, the procedure advances directly to step S6. After the procedure defined for step S6 is run and completed, the main loop returns to step S2 and repeats.
- It should be noted that the tasks that might be performed as the “other menu process” selected in S5 are not unique to or directly related to the present invention, and further description thereof is thus omitted below. The image registration process (S3) and the similar image searching process (S4) of the present invention are described next in detail below.
- Image Registration Process
- The image registration process shown as step S3 in FIG. 4 is described next in detail below with reference to the flow chart in FIG. 5. This process extracts feature quantities from a user-specified image to be stored in the image database, and then stores the specified image and extracted features to the
image database 50. - This routine starts by reading the user-specified image to be stored to the image database50 (S31). A feature extraction process is then run (S32) to extract feature quantities from the specified image. In this preferred embodiment of the invention, the extracted feature quantities include color, shape, and texture, but the invention shall obviously not be limited thereto. The extracted feature quantities are then related* to the image data of the specified image (S33), and finally registered in the
image database 50 together with the related specified image data (S34). A file name, image size information, and other image attributes are also normally stored with or related to the image data. The procedure then returns to the main loop (step S6 in FIG. 4). - Similar Image Searching Processing
- (a) Similar Image Search Based on Common Feature Quantities (First Embodiment)
- According to first embodiment, common feature quantities are extracted from a plurality of key images. Upon evaluating a similarity of a target image to a key image, only these common feature quantities are used for calculating a degree of similarity of the target image to a key image. In other words, feature quantities other than the common feature quantities are not used for calculation of the degree of similarity.
- The similar image searching processing according to first embodiment (shown as
step 4 in FIG. 4) is described below with reference to a flow chart shown in FIG. 6. In the present processing, predetermined feature quantities are extracted from each of plural key images specified by a user and feature quantities indicative of features common to the plural key images are selected as common feature quantities among the extracted feature quantities. Then, the searching processing is performed to find out an image similar to a key image among images registered in theimage database 50 based on the common feature quantities. - Referring to FIG. 6, at first, a plurality of key images specified by a user are read (step S41). Feature quantities of each key image are extracted from the image database 50 (step S42). In this step, feature quantities of plural kinds such as color, shape, texture and so on are obtained from each key image. Next, a search feature quantity compilation processing is performed to calculate common feature quantities for searching a similar image by comparing the feature quantities of respective key images with each other (step S43). This search feature quantity compilation processing will be described later.
- Thereafter, to find an image among images registered in the
image database 50 which is similar to the key images, the common feature quantities calculated in step S43 are compared with feature quantities of each image registered in the image database, and the degree of similarity between the key images and the presently selected image is calculated (from step S44 to step S46). - Namely, it is determined if there is an image in the
image database 50 for which the degree of similarity to the key images has not yet been calculated (herein below referred to as a “target image”) (step S44). If such a target image is present in theimage database 50, the feature quantities for that target image are read from the image database 50 (step S45). Then, a degree of similarity indicative of a similarity between the target image and each key image is calculated based on the common feature quantities of the key images and the feature quantities of the key images and the feature quantities of the target image (step S46) and the processing loops back to step S43. This calculation processing of the degree of similarity will be described later. These steps S43 to S45 are repeated until the similarity calculation has been completed for all target images in the image database. - When the degree of similarity has been calculated for all target images in the
image database 50, an output processing of search results is performed (step S47) and the processing returns to the main routine. In the output processing of the search results, it is determined that those target images for which the calculated degree of similarity exceeds a predetermined threshold value are similar to the key images. The target images determined to be similar are then displayed on screen in turn from the highest degree of similarity. - Alternately, without providing any threshold for determination of the similarity, a predetermined number of target images can be displayed in a descending order from the highest degree of similarity. Further, the search results can be displayed by simply outputting the file name of the target image data, preferably with related attributes such as its location in the database. The target images tested by the search loop from step S44 to step S46 shall also not be limited to all of the images stored in the
image database 50. For instance, only images preselected based on predetermined conditions can be tested. - The search feature quantity compilation processing (step S43 in FIG. 6) is described next with reference to a flow chart shown in FIG. 7. This processing calculates the common feature quantities to be used upon searching a target image which is similar in common to a plurality of key images. That is, the common feature quantities indicative of features in common to a plurality of key images are obtained by comparing feature quantities of the same kind among the plurality of key images and choosing feature quantities of the same kind having values being approximate to each other as the common feature quantities. This processing is described in further detail below.
- As shown in FIG. 7, the processing starts by calculating, for every kind of the feature quantity, differences among the feature quantities of the key images (step S431). Next, a mean value of the differences obtained in step S431 is calculated for every kind of the feature quantity (S432). An example of this calculation is shown in FIG. 8. This figure shows an example obtained for three
key images 1 to 3 each having fivefeature quantities 1 to 5 as shown in FIG. 9. The mean value of the differences with respect to the same feature quantity indicates the smaller the mean value the greater the image similarity with respect to that feature quantity. - Thereafter, every feature quantity or the kind of the feature quantity having a mean value equal to or less than a predetermined threshold is selected as a common feature quantity (step S433). That is, when the mean feature quantity difference is less than the threshold value, that feature quantity is determined to be common to the plural key images. In the example shown in FIG. 8, this threshold value is set to 0.1 and
feature quantity 2 andfeature quantity 3 are thus selected as common feature quantities. Thus, the feature quantities or kinds of feature quantities to be used for searching are determined by comparing feature quantities of the same kind in the key images. In the last step, a specific flag is set for every feature quantity selected as a common feature quantity (step S434). - Next, the processing for calculating the degree of similarity (step S46 in FIG. 6) is described below. Based on the common feature quantities extracted from the key images and a target image, this processing calculates the degree of similarity of the target image to all key images or either one of chem.
- According to the present embodiment, only feature quantities selected as common feature quantities in step S43 are used for calculating the degree of similarity of the target image to the key image.
- Accordingly, feature quantities other than common feature quantities are not used for the calculation of the degree of similarity. Due to this, images having features common to plural key images can be obtained as the results of search with a high accuracy. In this case, the target image similar to all of the plural key images. In other words, this search method imposes a search under an AND condition among the key images and thereby, enables to restrict ranges of searching conditions to reasonable ones.
- FIG. 11 shows an example of the calculation for the degree of similarity. In this example, the distance between each key image and the target image is calculated based on the values of
feature quantity 2 andfeature quantity 3 selected as the common feature quantities and the degree of similarity is obtained based on the calculated distances. The values shown in FIG. 11 are obtained from the following equations (1) to (3). - Difference between feature quantities=|(key image feature quantity)−(target image feature quantity)| (1)
- Distance=square root of sum of (each feature quantity difference )2 (2)
- Degree of similarity=1.0/distance (3)
- The degree of similarity thus obtained indicates that the image similarity becomes high as the calculated value of the degree of similarity becomes large.
- (b) Similar Image Searching Based on All Kinds of Feature Quantities (Second Embodiment)
- According to second embodiment, the degree of similarity between a target image and key images is calculated for each of the key images based on feature quantities of all kinds. From the degrees of similarity calculated for each key image, the highest degree of similarity is then selected as the degree of similarity of the target image to the corresponding key image. In this case, images similar to at least one of the plural key images are searched. That is, the image search is performed under an OR logic condition and, accordingly, it becomes possible to broaden the ranges of searching conditions by increasing the number of key images. Second embodiment performs a processing shown in FIG. 12 and FIG. 13 instead of FIGS. 6 and 7 of first embodiment.
- Referring to FIG. 12, a plurality of key images specified by a user are read at first (S141). Feature quantities are obtained from the
image database 50 for each key image (S142). Namely, feature quantities of plural kinds such as color, shape, texture and the like are obtained form each key image. - Then, the feature quantities of each key image obtained in step S142 are compared with feature quantities of each target image registered in the
image database 50, the similarity between each key image and the target image is determined and thereby, target images being similar to respective key images are searched (steps from S143 to S145). - At first, it is determined whether or not there is a target image in the image database for which the degree of similarity to each key image has not yet calculated (step S143). If such a target image is present in the
image database 50, the feature quantities for that target image are read from the image database (step S144). Then, calculation of the degree of similarity indicative of similarity between images is performed based on the feature quantities of each key image and those of the target image (step S144). After this calculation, the processing loops back to step S143. This calculation processing of the degree of similarity (step S145) will be described later. These steps S143 to S145 are repeated until the similarity calculation has been completed for all target images in theimage database 50. - When the degree of similarity has been calculated for all target images in the
image database 50, an output processing of search results is performed (step S147) and then, the processing returns to the main routine. In this output processing of search results (step S146), based on the degrees of similarity calculated for all target images in theimage database 50, the key image having a degree of similarity higher than a predetermined threshold value is determined to be similar to the corresponding key image. The target images determined to be similar are then display on screen in sequence from the highest degree of similarity. - Alternately, without providing any threshold for determination of the similarity, a predetermined number of target images can be displayed in a descending order from the highest degree of similarity. Further, the search results can be displayed by simply outputting the file name of the target image data, preferably with related attributes such as its location in the database. The target images tested by the search loop from step S144 to step S146 shall also not be limited to all of the images stored in the
image database 50. For instance, only images preselected based on predetermined conditions can be tested. - Next, the calculation processing of degree of similarity is described below with reference to a flow chart shown in FIG. 13.
- As shown in FIG. 14, a distance between a feature quantity of the key image and that of the same kind of the target image is calculated for each of all feature quantities, and a distance between the key images and the target image is calculated based on the distances of respective feature quantities.
- Based on the distance between images, the degree of similarity is calculated (step S1451). More specifically, the values shown in FIG. 14 are obtained from the following equations (4) to (6).
- Feature quantity distance=|(key image feature quantity)−(target image feature quantity)| (4)
- Distance=sqrt of {sum of (each feature quantity distance)2} (5)
- Degree of similarity=1.0/distance
- Note that with this method image similarity increases as the calculated degree of similarity value rises. Because the degree of similarity to
key image 3 is highest in the example shown in FIG. 14, that value is used as the degree of similarity between the current target image and the key images (step S1452). - The output processing of search results mentioned above with respect to step S146 is performed based on the degrees of similarity determined in step S1452.
- (c) Similar Image Searching Processing Using Degrees of Similarity Based on All Feature Quantities and Degrees of Similarity Based on Common Feature Quantities (Third Embodiment)
- This method combines the degrees of similarity based on the common feature quantities calculated according to first embodiment (a) and those based on all feature quantities calculated according to second embodiment (b), to obtain final degrees of similarity. In this case, the degree of similarity based on common feature quantities is weighted and the weighted degree of similarity and the degree of similarity based on all feature quantities are added to obtain a mean value thereof. This mean value is assigned as the degree of similarity of the target image to the key image. Thus, this method performs a similar image search in combination of AND and OR conditions. As a result, it becomes possible to find an image having features common to all key images among images similar to any one of the key images and therefore, to improve the search accuracy compared with second embodiment (b).
- An example of the calculation of the degree of similarity is shown in FIG. 15. In this example, the weight for the degree of similarity a calculated in FIG. 11 is set larger than that for the degree of similarity calculated in FIG. 14. A weighted mean value is than calculated to obtain the final degree of similarity c. This final degree of similarity c is obtained according to the next equation (7).
- Degree of similarity={(degree of similarity a×weight wa)+(degree of similarity b×weight wb)}/(wa+wb) (7)
- As stated above, according to the present embodiment, feature quantities are extracted from a plurality of key images, when specified and those common to the plural key images are selected to determine similarity between images using these common feature quantities. Thus, the common feature quantities are automatically determined from among plural key images and accordingly, the user burden can be reduced. It is also possible to search for images having features common to plural key images. This enables to narrow the ranges of respective searching conditions by specifying plural key images. Also, the search can be similarly expanded to broader ranges of searching conditions by considering the degree of similarity indicating similarity of an image to at least one of plural key images. Search accuracy can be thus improved.
- As will be known from the above descriptions of the present invention, a first image searching system, a first image searching method, and a first recording medium storing an image searching program embodying the first image searching method of the invention determines the feature quantities to be used for image searching based on the feature quantities extracted from one or more key images, and calculates a degree of similarity using only the feature quantities selected as search criteria. It is therefore possible to search for images with features common to those of the selected key images. The search conditions can thus be restricted by specifying plural key images, and search accuracy can thus be improved. The user burden is also reduced because feature quantities are automatically extracted from the specified key images, and similar images are searched for based on the extracted feature quantities.
- As will be known from the above descriptions of the present invention, a second image searching system, a second image searching method, and a second recording medium storing an image searching program embodying the second image searching method of the invention calculates the degree of similarity used for searching by combining a degree of similarity indicative of similarity to all of the plurality of key image, and a degree of similarity indicative of similarity to any one of the plurality of key image. It is therefore possible to set a wide range of search criteria by selecting a plurality of key images, and further improve the accuracy of similar image searches.
- Although the present invention has been described in connection with the preferred embodiments thereof with reference to the accompanying drawings, it is to be noted that various changes and modifications will be apparent to those skilled in the art. Such changes and modifications are to be understood as included within the scope of the present invention as defined by the appended claims, unless they depart therefrom.
Claims (30)
1. An image searching system comprising:
an image database storing a plurality of database images to be searched for; a specifying controller for specifying a plurality of key images used to specify search conditions;
an extracting controller for extracting common feature values of common images from the plural key images specified by the specifying controller;
a calculating controller for comparing the common feature values, extracted by the extracting controller, with the feature values of the plural database images to thereby sequentially calculate similarities between the common feature values and the database image feature values; and
a searching controller for retrieving from the database images one of the images which is similar to the key image, based on a similarity calculated by the calculating controller.
2. The system as claimed in claim 1 , wherein the extracting controller includes:
an extracting means for extracting a plurality of types of the feature quantities from the respective key images specified by the specifying controller;
a selecting means for comparing the feature quantities, extracted by the extracting means, among the plural key images specified by the specifying controller to thereby select at least one of the types of the feature quantities; and
a determining means for determining the common feature quantities based on the at least one type of the features quantities selected by the selecting means.
3. The system as claimed in claim 2 , wherein the selecting means is operable to compare the feature quantities of the same types among the plural key images specified by the specifying controller and wherein the determining means is operable to calculate an average value of the feature quantities of the plural key image with respect to the types of the feature quantities selected by the selecting means, to thereby determine the calculated average value as representing the common feature quantities.
4. An image searching system which comprises:
an image database storing a plurality of database images to be searched for;
a specifying controller for specifying a plurality of key images used to specify search conditions;
a calculating controller for comparing the plural images, specified by the specifying controller, with the plural database images to thereby calculate similarities between the common feature values and the database image feature values;
a selecting controller for retrieving a particular key image from the specified images based on the similarities calculated by the calculating controller; and
a searching controller for retrieving the images from the database images based on the similarity between the key image, selected by the selecting controller, and the database images.
5. The system as claimed in claim 4 , wherein the selecting controller is operable to select as a particular one of the plural specified key images, the key images which most resemble to the database images being searched for.
6. The system as claimed in claim 5 , wherein the calculating controller is operable to calculate a plurality of types of the feature quantities from the plural key images and then to calculate a degree of similarity by comparing the feature quantities with the database images for each type, and wherein the selecting controller selects, as the particular key image from the plural specified images, the key images which most resemble to the database images being searched with respect to an average value of degrees of similarities calculated by the calculating means for each type of the feature quantities.
7. An image searching system which comprises:
an image database storing a plurality of database images to be searched for;
a specifying controller for specifying a plurality of key images used to specify search conditions;
a first calculating controller for comparing all of the key images, specified by the specifying controller, with respective feature values of the database images to thereby calculate similarities therebetween;
a second calculating controller for selecting a particular key image from the plural key images specified by the specifying controller and for comparing the particular key image with the database images to thereby calculate similarities therebetween;
a third calculating controller for calculating a final similarity for use in searching based on the similarities calculated respectively by the first and second calculating controllers; and
a searching controller for retrieving one of the database images, which is similar to the particular key image, based on the final similarity calculated by the third calculating controller.
8. The image searching system as claimed in claim 7 , wherein the third calculating controller is operable to increase a weight of the degree of similarity, calculated by the first calculating controller, to a value greater than that of the degree of similarity, calculated by the second calculating controller, to thereby calculate the final degree of similarity.
9. The image searching system as claimed in claim 8 , wherein the first calculating controller is operable to extract the common feature quantities of the image common to all of the key images, and to compare those common feature quantities with the database image to thereby calculate the degree of similarity.
10. The image searching system as claimed in claim 9 , wherein the second calculating controller is operable to select the key images most similar to the database image from the key images and to calculate the degree of similarity.
11. An image searching method which comprises the steps of:
storing a plurality of database images to be searched for in a database;
specifying a plurality of key images used to specify search conditions by means of a specifying controller;
extracting by means of an extracting controller, common feature values of common images from the plural key images specified by the specifying controller;
comparing by means of a calculating controller, the common feature values, extracted by the extracting controller, with the feature values of the plural database images to thereby sequentially calculate similarities between the common feature values and the database image feature values; and
retrieving from the database images one of the images which is similar to the key image, based on a similarity calculated by the calculating controller, by means of a searching controller.
12. The method as claimed in claim 11 , wherein the extracting step includes the sub-steps of extracting a plurality of types of the feature quantities from the respective key images specified by the specifying controller; comparing by means of a selecting means the feature quantities, extracted by the extracting sub-step, among the plural key images specified by the specifying controller to thereby select at least one of the types of the feature quantities; and determining the common feature quantities based on the at least one type of the features quantities selected by the selecting means.
13. The method as claimed in claim 12 , wherein the selecting means is operable to compare the feature quantities of the same types among the plural key images specified by the specifying controller and wherein the determining means is operable to calculate an average value of the feature quantities of the plural key image with respect to the types of the feature quantities selected by the selecting means, to thereby determine the calculated average value as representing the common feature quantities.
14. An image searching method which comprises the steps of:
storing a plurality of database images to be searched for in an image database;
specifying a plurality of key images used to specify search conditions by means of a specifying controller;
comparing by means of a calculating controller, the plural images, specified by the specifying controller, with the plural database images to thereby calculate similarities between the common feature values and the database image feature values;
retrieving a particular key image from the specified images based on the similarities calculated by the calculating controller, by means of a selecting controller; and
retrieving the images from the database images based on the similarity between the key image, selected by the selecting controller, and the database images, by means of a searching controller.
15. The method as claimed in claim 14 , wherein the selecting controller is operable to select as a particular one of the plural specified key images, the key images which most resemble to the database images being searched for.
16. The method as claimed in claim 15 , wherein the calculating controller is operable to calculate a plurality of types of the feature quantities from the plural key images and then to calculate a degree of similarity by comparing the feature quantities with the database images for each type, and wherein the selecting controller selects, as the particular key image from the plural specified images, the key images which most resemble to the database images being searched with respect to an average value of degrees of similarities calculated by the calculating means for each type of the feature quantities.
17. An image searching method which comprises the steps of:
storing a plurality of database images to be searched for in an image database;
specifying a plurality of key images used to specify search conditions by means of a specifying controller;
comparing all of the key images, specified by the specifying controller, with respective feature values of the database images to thereby calculate similarities therebetween, by means of a first calculating controller;
selecting by means of a second calculating controller, a particular key image from the plural key images specified by the specifying controller and for comparing the particular key image with the database images to thereby calculate similarities therebetween;
calculating, by means of a third calculating controller, a final similarity for use in searching based on the similarities calculated respectively by the first and second calculating controllers; and
retrieving by means of a searching controller, one of the database images, which is similar to the particular key image, based on the final similarity calculated by the third calculating controller.
18. The image searching method as claimed in claim 17 , wherein the third calculating controller is operable to increase a weight of the degree of similarity, calculated by the first calculating controller, to a value greater than that of the degree of similarity, calculated by the second calculating controller, to thereby calculate the final degree of similarity.
19. The image searching method as claimed in claim 18 , wherein the first calculating controller is operable to extract the common feature quantities of the image common to all of the key images, and to compare those common feature quantities with the database image to thereby calculate the degree of similarity.
20. The image searching method as claimed in claim 19 , wherein the second calculating controller is operable to select the key images most similar to the database image from the key images and to calculate the degree of similarity.
21. A recording medium storing therein a computer-executable image searching program, said program comprising:
storing a plurality of database images to be searched for in a database;
specifying a plurality of key images used to specify search conditions by means of a specifying controller;
extracting by means of an extracting controller, common feature values of common images from the plural key images specified by the specifying controller;
comparing by means of a calculating controller, the common feature values, extracted by the extracting controller, with the feature values of the plural database images to thereby sequentially calculate similarities between the common feature values and the database image feature values; and
retrieving from the database images one of the images which is similar to the key image, based on a similarity calculated by the calculating controller, by means of a searching controller.
22. The recording medium as claimed in claim 21 , wherein the extracting controller includes an extracting means for extracting a plurality of types of the feature quantities from the respective key images specified by the specifying controller; a selecting means for comparing the feature quantities, extracted by the extracting sub-step, among the plural key images specified by the specifying controller to thereby select at least one of the types of the feature quantities; and a determining means for determining the common feature quantities based on the at least one type of the features quantities selected by the selecting means.
23. The recording medium as claimed in claim 22 , wherein the selecting means is operable to compare the feature quantities of the same types among the plural key images specified by the specifying controller and wherein the determining means is operable to calculate an average value of the feature quantities of the plural key image with respect to the types of the feature quantities selected by the selecting means, to thereby determine the calculated average value as representing the common feature quantities.
24. A recording medium storing therein a computer-executable image searching program, said program comprising:
storing a plurality of database images to be searched for in an image database;
specifying a plurality of key images used to specify search conditions by means of a specifying controller;
comparing by means of a calculating controller, the plural images, specified by the specifying controller, with the plural database images to thereby calculate similarities between the common feature values and the database image feature values;
retrieving a particular key image from the specified images based on the similarities calculated by the calculating controller, by means of a selecting controller; and
retrieving the images from the database images based on the similarity between the key image, selected by the selecting controller, and the database images, by means of a searching controller.
25. The recording medium as claimed in claim 24 , wherein the selecting controller is operable to select as a particular one of the plural specified key images, the key images which most resemble to the database images being searched for.
26. The recording medium as claimed in claim 25 , wherein the calculating controller is operable to calculate a plurality of types of the feature quantities from the plural key images and then to calculate a degree of similarity by comparing the feature quantities with the database images for each type, and wherein the selecting controller selects, as the particular key image from the plural specified images, the key images which most resemble to the database images being searched with respect to an average value of degrees of similarities calculated by the calculating means for each type of the feature quantities.
27. A recording medium storing therein a computer-executable image searching program, said program comprising:
storing a plurality of database images to be searched for in an image database;
specifying a plurality of key images used to specify search conditions by means of a specifying controller;
comparing all of the key images, specified by the specifying controller, with respective feature values of the database images to thereby calculate similarities therebetween, by means of a first calculating controller;
selecting by means of a second calculating controller, a particular key image from the plural key images specified by the specifying controller and for comparing the particular key image with the database images to thereby calculate similarities therebetween;
calculating, by means of a third calculating controller, a final similarity for use in searching based on the similarities calculated respectively by the first and second calculating controllers; and
retrieving by means of a searching controller, one of the database images, which is similar to the particular key image, based on the final similarity calculated by the third calculating controller.
28. The recording medium as claimed in claim 27 , wherein the third calculating controller is operable to increase a weight of the degree of similarity, calculated by the first calculating controller, to a value greater than that of the degree of similarity, calculated by the second calculating controller, to thereby calculate the final degree of similarity.
29. The recording medium as claimed in claim 28 , wherein the first calculating controller is operable to extract the common feature quantities of the image common to all of the key images, and to compare those common feature quantities with the database image to thereby calculate the degree of similarity.
30. The recording medium as claimed in claim 29 , wherein the second calculating controller is operable to select the key images most similar to the database image from the key images and to calculate the degree of similarity.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP04198998A JP4074366B2 (en) | 1998-02-24 | 1998-02-24 | Image search apparatus and method, and recording medium storing image search program |
JP10-41989 | 1998-02-24 |
Publications (2)
Publication Number | Publication Date |
---|---|
US20020178135A1 true US20020178135A1 (en) | 2002-11-28 |
US7392233B2 US7392233B2 (en) | 2008-06-24 |
Family
ID=12623614
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US09/255,352 Expired - Fee Related US7392233B2 (en) | 1998-02-24 | 1999-02-23 | Image searching system, image searching method, and a recording medium storing an image searching program |
Country Status (2)
Country | Link |
---|---|
US (1) | US7392233B2 (en) |
JP (1) | JP4074366B2 (en) |
Cited By (32)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020009231A1 (en) * | 2000-03-29 | 2002-01-24 | Kunihiro Yamamoto | Image retrieval method and apparatus |
US6618501B1 (en) * | 1999-05-06 | 2003-09-09 | Canon Kabushiki Kaisha | Object similarity calculation method and apparatus |
US20040003001A1 (en) * | 2002-04-03 | 2004-01-01 | Fuji Photo Film Co., Ltd. | Similar image search system |
US20040220962A1 (en) * | 2003-04-30 | 2004-11-04 | Canon Kabushiki Kaisha | Image processing apparatus, method, storage medium and program |
US20050073721A1 (en) * | 2003-10-06 | 2005-04-07 | Canon Kabushiki Kaisha | Form recognizing apparatus, form recognizing method, program and storage medium |
EP1538822A1 (en) * | 2003-12-04 | 2005-06-08 | Konica Minolta Holdings, Inc. | Image control apparatus, image control method and program |
US20050234719A1 (en) * | 2004-03-26 | 2005-10-20 | Tatsuya Hosoda | Selection of images for image processing |
US7055168B1 (en) * | 2000-05-03 | 2006-05-30 | Sharp Laboratories Of America, Inc. | Method for interpreting and executing user preferences of audiovisual information |
US20060218522A1 (en) * | 2005-03-25 | 2006-09-28 | Vistaprint Technologies Limited | Selecting images using associated keywords |
US20070031035A1 (en) * | 2005-08-04 | 2007-02-08 | Gunnar Hovden | System and method for effectively implementing a texture feature detector |
US20070185837A1 (en) * | 2006-02-09 | 2007-08-09 | Microsoft Corporation | Detection of lists in vector graphics documents |
US20080240575A1 (en) * | 2007-03-30 | 2008-10-02 | Navneet Panda | Learning concept templates from web images to query personal image databases |
US20080263036A1 (en) * | 2006-12-13 | 2008-10-23 | Canon Kabushiki Kaisha | Document search apparatus, document search method, program, and storage medium |
US20090132523A1 (en) * | 2007-11-16 | 2009-05-21 | Canon Kabushiki Kaisha | Information processing apparatus, control method thereof, computer program, and storage medium |
US7653131B2 (en) | 2001-10-19 | 2010-01-26 | Sharp Laboratories Of America, Inc. | Identification of replay segments |
US7657907B2 (en) | 2002-09-30 | 2010-02-02 | Sharp Laboratories Of America, Inc. | Automatic user profiling |
US7793205B2 (en) | 2002-03-19 | 2010-09-07 | Sharp Laboratories Of America, Inc. | Synchronization of video and data |
US7904814B2 (en) | 2001-04-19 | 2011-03-08 | Sharp Laboratories Of America, Inc. | System for presenting audio-video content |
US7961938B1 (en) * | 2006-06-30 | 2011-06-14 | Adobe Systems Incorporated | Finding and structuring images based on a color search |
US8020183B2 (en) | 2000-09-14 | 2011-09-13 | Sharp Laboratories Of America, Inc. | Audiovisual management system |
US8028314B1 (en) | 2000-05-26 | 2011-09-27 | Sharp Laboratories Of America, Inc. | Audiovisual information management system |
US8356317B2 (en) | 2004-03-04 | 2013-01-15 | Sharp Laboratories Of America, Inc. | Presence based technology |
US20130074034A1 (en) * | 2010-06-02 | 2013-03-21 | Allen Learning Technologies | Logic table |
US8689253B2 (en) | 2006-03-03 | 2014-04-01 | Sharp Laboratories Of America, Inc. | Method and system for configuring media-playing sets |
US8776142B2 (en) | 2004-03-04 | 2014-07-08 | Sharp Laboratories Of America, Inc. | Networked video devices |
US8949899B2 (en) | 2005-03-04 | 2015-02-03 | Sharp Laboratories Of America, Inc. | Collaborative recommendation system |
US20150035827A1 (en) * | 2012-03-29 | 2015-02-05 | Sony Corporation | Information processing device, information processing method, and information processing system |
US20150043828A1 (en) * | 2013-08-08 | 2015-02-12 | Stmicroelectronics Sa | Method for searching for a similar image in an image database based on a reference image |
US20150071529A1 (en) * | 2013-09-12 | 2015-03-12 | Kabushiki Kaisha Toshiba | Learning image collection apparatus, learning apparatus, and target object detection apparatus |
EP2980751A4 (en) * | 2013-03-28 | 2016-08-31 | Fujifilm Corp | Image retrieval device, operation control method therefor, and image retrieval server |
US20180157682A1 (en) * | 2015-06-10 | 2018-06-07 | We'll Corporation | Image information processing system |
CN111241332A (en) * | 2020-01-17 | 2020-06-05 | 珠海博明视觉科技有限公司 | Method for searching corresponding measuring program through automatic identification of workpiece |
Families Citing this family (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AUPQ589300A0 (en) * | 2000-02-28 | 2000-03-23 | Canon Kabushiki Kaisha | Automatically selecting fonts |
US7620622B1 (en) * | 2001-03-08 | 2009-11-17 | Yahoo! Inc. | Method and system for indexing information and providing results for a search including objects having predetermined attributes |
JP4363792B2 (en) | 2001-03-23 | 2009-11-11 | 富士通株式会社 | Information retrieval system and method |
US8022980B2 (en) * | 2002-02-12 | 2011-09-20 | Given Imaging Ltd. | System and method for displaying an image stream |
US7474327B2 (en) * | 2002-02-12 | 2009-01-06 | Given Imaging Ltd. | System and method for displaying an image stream |
JP4082043B2 (en) * | 2002-02-27 | 2008-04-30 | 富士ゼロックス株式会社 | Image search device |
JP5248780B2 (en) * | 2003-12-31 | 2013-07-31 | ギブン イメージング リミテッド | System and method for displaying an image stream |
JP2006040085A (en) | 2004-07-29 | 2006-02-09 | Sony Corp | Information processing device and method therefor, storage medium, and program |
JP2006285526A (en) * | 2005-03-31 | 2006-10-19 | Seiko Epson Corp | Information retrieval according to image data |
JP4706415B2 (en) * | 2005-09-27 | 2011-06-22 | カシオ計算機株式会社 | Imaging apparatus, image recording apparatus, and program |
JP5237724B2 (en) * | 2008-08-19 | 2013-07-17 | オリンパスイメージング株式会社 | Image search system |
US8682142B1 (en) | 2010-03-18 | 2014-03-25 | Given Imaging Ltd. | System and method for editing an image stream captured in-vivo |
JP5261776B2 (en) * | 2010-04-21 | 2013-08-14 | 日本電信電話株式会社 | Similar data discriminating apparatus, similar data discriminating method and program |
WO2011135573A1 (en) | 2010-04-28 | 2011-11-03 | Given Imaging Ltd. | System and method for displaying portions of in-vivo images |
US9026034B2 (en) | 2010-05-04 | 2015-05-05 | Project Oda, Inc. | Automatic detection of broadcast programming |
JP2011257979A (en) * | 2010-06-09 | 2011-12-22 | Olympus Imaging Corp | Image retrieval device, image retrieval method, and camera |
JP5618787B2 (en) | 2010-11-29 | 2014-11-05 | キヤノン株式会社 | Report creation support apparatus, creation support method thereof, and program |
WO2013012952A1 (en) | 2011-07-18 | 2013-01-24 | Viggle Inc | System and method for tracking and rewarding media and entertainment usage including substanitally real time rewards |
JP5794185B2 (en) * | 2012-03-21 | 2015-10-14 | カシオ計算機株式会社 | Image processing apparatus, image processing method, and program |
JP5552552B2 (en) * | 2013-03-22 | 2014-07-16 | オリンパスイメージング株式会社 | Image search system |
JP5863916B2 (en) * | 2014-09-18 | 2016-02-17 | キヤノン株式会社 | Report creation support apparatus, creation support method thereof, and program |
US11573706B2 (en) | 2021-06-07 | 2023-02-07 | Western Digital Technologies, Inc. | Data storage device and method for efficient image searching |
Family Cites Families (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3177746B2 (en) * | 1991-03-20 | 2001-06-18 | 株式会社日立製作所 | Data processing system and method |
JPH05242160A (en) | 1992-02-27 | 1993-09-21 | Matsushita Electric Ind Co Ltd | Image feature extraction device, image collation device, and image retrieval device |
US5586240A (en) * | 1992-03-11 | 1996-12-17 | Genesis Software, Inc. | Image generation and retrieval system integrated with arbitrary application using layered interface |
JPH0765169A (en) | 1993-08-30 | 1995-03-10 | Nippon Telegr & Teleph Corp <Ntt> | Image retrieving method |
JPH07131641A (en) * | 1993-11-08 | 1995-05-19 | Canon Inc | Image processor |
JPH07234877A (en) | 1994-02-23 | 1995-09-05 | Nippon Telegr & Teleph Corp <Ntt> | Data base retrieving method |
US6246804B1 (en) * | 1994-11-15 | 2001-06-12 | Canon Kabushiki Kaisha | Image retrieval method and apparatus using a compound image formed from a plurality of detected regions |
JPH08329099A (en) | 1995-05-30 | 1996-12-13 | Minolta Co Ltd | Image information processor |
JPH08329096A (en) | 1995-05-30 | 1996-12-13 | Minolta Co Ltd | Image data retrieval device |
DE69628282T2 (en) * | 1995-09-15 | 2004-03-11 | Interval Research Corp., Palo Alto | METHOD FOR COMPRESSING SEVERAL VIDEO IMAGES |
JPH09101970A (en) * | 1995-10-06 | 1997-04-15 | Omron Corp | Method and device for retrieving image |
JPH09114851A (en) * | 1995-10-20 | 1997-05-02 | Fuji Xerox Co Ltd | Information managing device |
WO1998035468A2 (en) * | 1997-01-27 | 1998-08-13 | Benjamin Slotznick | System for delivering and displaying primary and secondary information |
-
1998
- 1998-02-24 JP JP04198998A patent/JP4074366B2/en not_active Expired - Fee Related
-
1999
- 1999-02-23 US US09/255,352 patent/US7392233B2/en not_active Expired - Fee Related
Cited By (52)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6618501B1 (en) * | 1999-05-06 | 2003-09-09 | Canon Kabushiki Kaisha | Object similarity calculation method and apparatus |
US20020009231A1 (en) * | 2000-03-29 | 2002-01-24 | Kunihiro Yamamoto | Image retrieval method and apparatus |
US7055168B1 (en) * | 2000-05-03 | 2006-05-30 | Sharp Laboratories Of America, Inc. | Method for interpreting and executing user preferences of audiovisual information |
US8028314B1 (en) | 2000-05-26 | 2011-09-27 | Sharp Laboratories Of America, Inc. | Audiovisual information management system |
US8020183B2 (en) | 2000-09-14 | 2011-09-13 | Sharp Laboratories Of America, Inc. | Audiovisual management system |
US7904814B2 (en) | 2001-04-19 | 2011-03-08 | Sharp Laboratories Of America, Inc. | System for presenting audio-video content |
US7653131B2 (en) | 2001-10-19 | 2010-01-26 | Sharp Laboratories Of America, Inc. | Identification of replay segments |
US7853865B2 (en) | 2002-03-19 | 2010-12-14 | Sharp Laboratories Of America, Inc. | Synchronization of video and data |
US8214741B2 (en) | 2002-03-19 | 2012-07-03 | Sharp Laboratories Of America, Inc. | Synchronization of video and data |
US7793205B2 (en) | 2002-03-19 | 2010-09-07 | Sharp Laboratories Of America, Inc. | Synchronization of video and data |
US7374077B2 (en) * | 2002-04-03 | 2008-05-20 | Fujifilm Corporation | Similar image search system |
US20040003001A1 (en) * | 2002-04-03 | 2004-01-01 | Fuji Photo Film Co., Ltd. | Similar image search system |
US7657907B2 (en) | 2002-09-30 | 2010-02-02 | Sharp Laboratories Of America, Inc. | Automatic user profiling |
US20040220962A1 (en) * | 2003-04-30 | 2004-11-04 | Canon Kabushiki Kaisha | Image processing apparatus, method, storage medium and program |
US7548916B2 (en) * | 2003-04-30 | 2009-06-16 | Canon Kabushiki Kaisha | Calculating image similarity using extracted data |
US20050073721A1 (en) * | 2003-10-06 | 2005-04-07 | Canon Kabushiki Kaisha | Form recognizing apparatus, form recognizing method, program and storage medium |
US7450262B2 (en) * | 2003-10-06 | 2008-11-11 | Canon Kabushiki Kaisha | Form recognizing apparatus, form recognizing method, program and storage medium |
EP1538822A1 (en) * | 2003-12-04 | 2005-06-08 | Konica Minolta Holdings, Inc. | Image control apparatus, image control method and program |
US7515754B2 (en) | 2003-12-04 | 2009-04-07 | Konica Minolta Holdings, Inc. | Image control apparatus, image control method and program |
US8776142B2 (en) | 2004-03-04 | 2014-07-08 | Sharp Laboratories Of America, Inc. | Networked video devices |
US8356317B2 (en) | 2004-03-04 | 2013-01-15 | Sharp Laboratories Of America, Inc. | Presence based technology |
US7711210B2 (en) * | 2004-03-26 | 2010-05-04 | Seiko Epson Corporation | Selection of images for image processing |
US20050234719A1 (en) * | 2004-03-26 | 2005-10-20 | Tatsuya Hosoda | Selection of images for image processing |
US8949899B2 (en) | 2005-03-04 | 2015-02-03 | Sharp Laboratories Of America, Inc. | Collaborative recommendation system |
US8694494B2 (en) * | 2005-03-25 | 2014-04-08 | Vistaprint Schweiz Gmbh | Selecting images using associated keywords |
US20060218522A1 (en) * | 2005-03-25 | 2006-09-28 | Vistaprint Technologies Limited | Selecting images using associated keywords |
US20070031035A1 (en) * | 2005-08-04 | 2007-02-08 | Gunnar Hovden | System and method for effectively implementing a texture feature detector |
US7609885B2 (en) | 2005-08-04 | 2009-10-27 | Sony Corporation | System and method for effectively implementing a texture feature detector |
US7836399B2 (en) | 2006-02-09 | 2010-11-16 | Microsoft Corporation | Detection of lists in vector graphics documents |
US20070185837A1 (en) * | 2006-02-09 | 2007-08-09 | Microsoft Corporation | Detection of lists in vector graphics documents |
US8689253B2 (en) | 2006-03-03 | 2014-04-01 | Sharp Laboratories Of America, Inc. | Method and system for configuring media-playing sets |
US20110170770A1 (en) * | 2006-06-30 | 2011-07-14 | Adobe Systems Incorporated | Finding and structuring images based on a color search |
US7961938B1 (en) * | 2006-06-30 | 2011-06-14 | Adobe Systems Incorporated | Finding and structuring images based on a color search |
US8699826B2 (en) | 2006-06-30 | 2014-04-15 | Adobe Systems Incorporated | Finding and structuring images based on a color search |
US20080263036A1 (en) * | 2006-12-13 | 2008-10-23 | Canon Kabushiki Kaisha | Document search apparatus, document search method, program, and storage medium |
US20080240575A1 (en) * | 2007-03-30 | 2008-10-02 | Navneet Panda | Learning concept templates from web images to query personal image databases |
US8958661B2 (en) * | 2007-03-30 | 2015-02-17 | Intel Corporation | Learning concept templates from web images to query personal image databases |
US20090132523A1 (en) * | 2007-11-16 | 2009-05-21 | Canon Kabushiki Kaisha | Information processing apparatus, control method thereof, computer program, and storage medium |
US9507570B2 (en) * | 2010-06-02 | 2016-11-29 | Allen Learning Technologies | Method and program for creating applications by using a logic table |
US8914773B2 (en) * | 2010-06-02 | 2014-12-16 | Allen Learning Technologies | Logic table |
US20150074638A1 (en) * | 2010-06-02 | 2015-03-12 | Allen Learning Technologies | Logic table |
US20130074034A1 (en) * | 2010-06-02 | 2013-03-21 | Allen Learning Technologies | Logic table |
US9852358B2 (en) * | 2012-03-29 | 2017-12-26 | Sony Corporation | Information processing device, information processing method, and information processing system |
US20150035827A1 (en) * | 2012-03-29 | 2015-02-05 | Sony Corporation | Information processing device, information processing method, and information processing system |
EP2980751A4 (en) * | 2013-03-28 | 2016-08-31 | Fujifilm Corp | Image retrieval device, operation control method therefor, and image retrieval server |
EP3367338A1 (en) * | 2013-03-28 | 2018-08-29 | FUJIFILM Corporation | Image search apparatus, method of controlling operation of same, and image search server |
US9418313B2 (en) * | 2013-08-08 | 2016-08-16 | Stmicroelectronics Sa | Method for searching for a similar image in an image database based on a reference image |
US20150043828A1 (en) * | 2013-08-08 | 2015-02-12 | Stmicroelectronics Sa | Method for searching for a similar image in an image database based on a reference image |
US9158996B2 (en) * | 2013-09-12 | 2015-10-13 | Kabushiki Kaisha Toshiba | Learning image collection apparatus, learning apparatus, and target object detection apparatus |
US20150071529A1 (en) * | 2013-09-12 | 2015-03-12 | Kabushiki Kaisha Toshiba | Learning image collection apparatus, learning apparatus, and target object detection apparatus |
US20180157682A1 (en) * | 2015-06-10 | 2018-06-07 | We'll Corporation | Image information processing system |
CN111241332A (en) * | 2020-01-17 | 2020-06-05 | 珠海博明视觉科技有限公司 | Method for searching corresponding measuring program through automatic identification of workpiece |
Also Published As
Publication number | Publication date |
---|---|
JP4074366B2 (en) | 2008-04-09 |
US7392233B2 (en) | 2008-06-24 |
JPH11238078A (en) | 1999-08-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7392233B2 (en) | Image searching system, image searching method, and a recording medium storing an image searching program | |
US5644765A (en) | Image retrieving method and apparatus that calculates characteristic amounts of data correlated with and identifying an image | |
US7130849B2 (en) | Similarity-based search method by relevance feedback | |
US6549913B1 (en) | Method for compiling an image database, an image database system, and an image data storage medium | |
US6381365B2 (en) | Image data processing apparatus and image data processing method | |
US6853389B1 (en) | Information searching apparatus, information searching method, and storage medium | |
US6330576B1 (en) | User-friendly information processing device and method and computer program product for retrieving and displaying objects | |
US6804420B2 (en) | Information retrieving system and method | |
JPH1196364A (en) | Image feature amount comparing device and recording medium recording image feature amount comparison program | |
JPH1125113A (en) | Image retrieving device, generating method for key text for image retrieval, program for functioning computer as device therefor, and computer readable record medium recording program for executing method with computer | |
US7308119B2 (en) | Image retrieval apparatus and method, and image display apparatus and method thereof | |
JPH0830763A (en) | Method and device for processing multi-media data | |
JP2005236646A (en) | Image display device, method, and program | |
US6606636B1 (en) | Method and apparatus for retrieving dynamic images and method of and apparatus for managing images | |
JPH10162020A (en) | Browsing method for image data base | |
EP0619550B1 (en) | Apparatus and method for searching data | |
JPH1185982A (en) | Picture feature amount comparator and recording medium for recording picture feature amount comparison program | |
JPH11250077A (en) | Information processor, information processing method and record medium | |
JP2000048181A (en) | Image comparing device and method therefor, and recording medium | |
JPS6136868A (en) | Information retriever | |
JP2001134765A (en) | Method and device for retrieving image | |
JPH11238077A (en) | Image retrieval device and method and medium recording image retrieval program | |
JP3530993B2 (en) | Information processing apparatus and method | |
JPH11212993A (en) | Picture retrieval device/method and record medium recording picture retrieval program | |
JPH09204444A (en) | Information processing system and recording medium storing program for computer to perform processing for the same |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: MINOLTA CO., LTD., JAPAN Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:TANAKA, SUMIYO;REEL/FRAME:009787/0975 Effective date: 19990215 |
|
FEPP | Fee payment procedure |
Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
REMI | Maintenance fee reminder mailed | ||
LAPS | Lapse for failure to pay maintenance fees | ||
STCH | Information on status: patent discontinuation |
Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362 |
|
FP | Lapsed due to failure to pay maintenance fee |
Effective date: 20160624 |